DataScience-Instruct-500K

<p align="center" width="100%"> <img src="assets/logo.png" alt="DeepAnalyze" style="width: 60%; min-width: 300px; display: block; margin: auto;"> </p> # DeepAnalyze: Agentic Large Language Models for Autonomous Data Science [](https://arxiv.org/abs/2510.16872) [](https://github.com/ruc-datalab/DeepAnalyze) [](https://ruc-deepanalyze.github.io/) [](https://huggingface.co/RUC-DataLab/DeepAnalyze-8B) [](https://huggingface.co/datasets/RUC-DataLab/DataScience-Instruct-500K)  > **Authors**: **[Shaolei Zhang](https://zhangshaolei1998.github.io/), [Ju Fan*](http://iir.ruc.edu.cn/~fanj/), [Meihao Fan](https://scholar.google.com/citations?user=9RTm2qoAAAAJ), [Guoliang Li](https://dbgroup.cs.tsinghua.edu.cn/ligl/), [Xiaoyong Du](http://info.ruc.edu.cn/jsky/szdw/ajxjgcx/jsjkxyjsx1/js2/7374b0a3f58045fc9543703ccea2eb9c.htm)** **DeepAnalyze** is the first agentic LLM for autonomous data science. It can autonomously complete a wide range of data-centric tasks without human intervention, supporting: - 🛠 **Entire data science pipeline**: Automatically perform any data science tasks such as data preparation, analysis, modeling, visualization, and report generation. - 🔍 **Open-ended data research**: Conduct deep research on diverse data sources, including structured data (Databases, CSV, Excel), semi-structured data (JSON, XML, YAML), and unstructured data (TXT, Markdown), and finally produce analyst-grade research reports. - 📊 **Fully open-source**: The [model](https://huggingface.co/RUC-DataLab/DeepAnalyze-8B), [code](https://github.com/ruc-datalab/DeepAnalyze), [training data](https://huggingface.co/datasets/RUC-DataLab/DataScience-Instruct-500K), and [demo](https://huggingface.co/RUC-DataLab/DeepAnalyze-8B) of DeepAnalyze are all open-sourced, allowing you to deploy or extend your own data analysis assistant. <p align="center" width="100%"> <img src="./assets/deepanalyze.jpg" alt="deepanalyze" style="width: 70%; min-width: 300px; display: block; margin: auto;"> </p> More information refer to [DeepAnalyze's Repo](https://github.com/ruc-datalab/DeepAnalyze)<p align="center" width="100%"> <img src="assets/logo.png" alt="DeepAnalyze" style="width: 60%; min-width: 300px; display: block; margin: auto;"> </p> # DeepAnalyze：面向自主数据科学的智能体式大语言模型（Agentic Large Language Models） [](https://arxiv.org/abs/2510.16872) [](https://github.com/ruc-datalab/DeepAnalyze) [](https://ruc-deepanalyze.github.io/) [](https://huggingface.co/RUC-DataLab/DeepAnalyze-8B) [](https://huggingface.co/datasets/RUC-DataLab/DataScience-Instruct-500K)  > **作者**：**[张绍磊](https://zhangshaolei1998.github.io/)、[范举*](http://iir.ruc.edu.cn/~fanj/)、[范美浩](https://scholar.google.com/citations?user=9RTm2qoAAAAJ)、[李国良](https://dbgroup.cs.tsinghua.edu.cn/ligl/)、[杜小勇](http://info.ruc.edu.cn/jsky/szdw/ajxjgcx/jsjkxyjsx1/js2/7374b0a3f58045fc9543703ccea2eb9c.htm)** **DeepAnalyze** 是首款面向自主数据科学的智能体式大语言模型（Agentic Large Language Models）。它可在无需人工干预的前提下，自主完成各类以数据为中心的任务，支持以下能力： - 🛠 **完整数据科学流程**：自动执行各类数据科学任务，涵盖数据准备、数据分析、模型构建、可视化与报告生成。 - 🔍 **开放式数据研究**：对多类数据源开展深度研究，包括结构化数据（数据库、CSV、Excel）、半结构化数据（JSON、XML、YAML）以及非结构化数据（TXT、Markdown），最终产出分析师级别的专业研究报告。 - 📊 **完全开源**：DeepAnalyze的[模型](https://huggingface.co/RUC-DataLab/DeepAnalyze-8B)、[代码](https://github.com/ruc-datalab/DeepAnalyze)、[训练数据集](https://huggingface.co/datasets/RUC-DataLab/DataScience-Instruct-500K)与[演示程序](https://huggingface.co/RUC-DataLab/DeepAnalyze-8B)均已开源，支持用户自主部署或拓展自研数据分析助手。 <p align="center" width="100%"> <img src="./assets/deepanalyze.jpg" alt="DeepAnalyze" style="width: 70%; min-width: 300px; display: block; margin: auto;"> </p> 更多信息请参阅 [DeepAnalyze 代码仓库](https://github.com/ruc-datalab/DeepAnalyze)

搜集汇总

背景与挑战

背景概述

DataScience-Instruct-500K是一个用于自主数据科学任务的开源数据集，支持整个数据科学流程，包括数据准备、分析和建模等。数据集由RUC-DataLab发布，完全开源，可用于部署或扩展数据分析助手。